Psychosocial risk factors for neck pain: A

AMERICAN JOURNAL OF INDUSTRIAL MEDICINE 39:180±193 (2001)

Psychosocial Risk Factors for Neck Pain: A Systematic Review Èns, MSc,1,2,3 Willem van Mechelen, MD, PhD1,2, Paulien M. Bongers, Geertje A.M. Arie Lex M. Bouter, PhD,2 and Gerrit van der Wal, MD, PhD1,2

3 PhD,

Background Neck pain, which is assumed to be a multifactorial disease, is a major problem in modern society. Methods To identify the most important psychosocial risk factors for neck pain, a systematic review of the literature was carried out. The methodological quality of all studies in the review was assessed. Four levels of evidence were de®ned to assess the strength of evidence for potential risk factors for neck pain (strong, moderate, some or inconclusive evidence). Results Some evidence was found for a positive relationship between neck pain and high quantitative job demands, low social (coworker) support, low job control, high and low skill discretion and low job satisfaction. Inconclusive evidence was found for high job strain, low supervisor support, con¯icts at work, low job security, and limited rest break opportunities. Conclusion The procedure of the assessment of the methodological quality and the rating system applied to distinguish between high- and low-score studies, had a considerable in¯uence on the level of evidence, indicating that changes in this procedure may have a major impact on the overall conclusions of this review. Am. J. Ind. Med. 39:180±193, 2001. ß 2001 Wiley-Liss, Inc. KEY WORDS: neck pain; systematic review; psychosocial; risk factors; observational studies

INTRODUCTION Musculoskeletal problems are a major problem in modern society. In the Netherlands, the costs of work-related sick leave and medical consumption are very high (US$ 4.8 billion a year). Around 40% of these total costs is due to musculoskeletal disorders [Koningsveld and Mossink, 1997]. Neck pain may not be the biggest musculoskeletal problem, but it still is substantial. Recent prevalence data 1 Department of Social Medicine,Faculty of Medicine,Vrije Universiteit Amsterdam,van der Boechorststraat 7, 1081BTAmsterdam,The Netherlands 2 Institute for Research in Extramural Medicine, Faculty of Medicine,Vrije Universiteit Amsterdam, van der Boechorststraat 7, 1081BTAmsterdam,The Netherlands 3 TNO Work and Employment, P.O. Box 718, 2130 AS, Hoofddorp,The Netherlands Correspondence to: Geertje A.M. Arie«ns,TNO Work and Employment, P.O. Box 718, 2130 AS Hoofddorp,The Netherlands. E-mail: [email protected]

Accepted1September 2000

ß 2001Wiley-Liss, Inc.

showed that in a general population the 1-year prevalence of neck pain was 15% and 17% for males and females, respectively [Lau et al., 1996]. Prevalence data in occupational settings are even more impressive. Skov et al. [1996] reported 1-year prevalences of neck pain in a population of sales people (n 1304) of 54% in males and 76% in females. Neck pain is assumed to be a multifactorial disease, implying that there are a number of risk factors contributing to its development. Risk factors can be work-related, as well as non-work-related. Furthermore, risk factors can be divided roughly in three groups, i.e., physical, psychosocial, and individual-related risk factors. Many studies have been conducted to identify the most important risk factors for neck pain. Most studies focus on only one or a few risk factors, or on one particular category of risk factors. While most attention has always been given to physical risk factors for neck pain, psychosocial risk factors also seem to play a major role in the development of neck pain.

Psychosocial Risk Factors for Neck Pain

To identify the most important psychosocial risk factors for neck pain, a systematic review of the literature was carried out. This review deals with psychosocial risk factors at work and in leisure time, such as demands and control over work, work organization factors, work satisfaction, and social support at work and in leisure time. Individual-related psychological factors such as coping behavior are not within the scope of this review. A complementary systematic review concerning physical risk factors for neck pain has been published elsewhere [ArieÈns et al., 2000].

MATERIALS AND METHODS On-line searches in Medline, Embase, Psychlit, Sportdiscus, HSELINE, CISDOC, and NIOSHTIC were carried out to identify all relevant studies. The search concerned the time period January 1966 to November 1997, using the following keywords (MeSH and text words): neck, neck pain, risk factors, determinants, causality, work, exercise, overuse, physical load, workload, psychosocial factors. Abstracts of all studies identi®ed were read. If no abstract was available, or if, based on the abstract, it was unclear whether a study should enter this systematic review, the whole article was retrieved and read. In order to be included, a study had to meet the following criteria: 1. 2. 3. 4. 5.

6.

The study population must be a working population or a community-based population. Studies of patient populations were excluded. The study design must either be case±control, crosssectional, or cohort. The assessment of exposure should at least concern one psychosocial factor at work or during leisure time. The assessment of exposure may not be based just on job titles. The outcome can include one or more syndromes, signs or symptoms of the neck. The outcome can be a self-reported variable, as well as a clinical diagnosis. The outcome must be reported for the neck region separately. The study must be a full, peer-reviewed report published in the English, Dutch or German language.

Reference lists of included studies were checked for additional references. To check the selection procedure, a random sample of 30 studies was judged by a second reviewer to determine whether or not a study should be included in the review. The methodological quality of all studies that entered the review was assessed by means of a methodological quality assessment list. After critically reviewing existing quality lists [Stock, 1991; van Tulder et al., 1997; Borghouts et al., 1998], a criteria list was developed to assess the methodological quality of observational studies. The criteria

181

list contained various items on information and validity and/ or precision in ®ve categories: study purpose, study design, exposure measurements, outcome measurements, and analysis and data presentation. Separate quality assessment lists were constructed for cross-sectional, case±control, and cohort studies. In Table I the items of the methodological quality assessment lists are presented. The ®rst two items of the quality assessment list (items A and B) provide descriptive information only, while all other items of the list concern the validity and/or precision of the study (see also Appendix A). For every item in the quality list, two independent reviewers (GA and WvM) rated each study either ``positive'' (), ``negative'' (ÿ), or `ùnclear'' (?) if a study did or did not meet an item, or if no clear information was stated regarding that item, respectively. Results of these two independent reviewers were compared and, if differing, in a meeting consensus upon each item was reached. For each study, a total quality score was calculated by counting the number of validity/precision items that were rated positively. Based on this total score, a study was either categorized as high or low-score study. A high-score was arbitrarily de®ned as a study that scored positively on at least 50% of the validity/precision items of the methodological quality list concerned. Low-score studies scored positively on less than 50% of the validity/precision items. The strength of evidence for potential risk factors was assessed by de®ning four levels of evidence as follows: 1. 2. 3.

4.

Strong evidence: consistent ®ndings in multiple highscore cohort and/or case±control studies. Moderate evidence: consistent ®ndings in multiple cohort and/or case±control studies, of which only one study is a high-score study. Some evidence: ®ndings of one cohort or case±control study, or consistent ®ndings in multiple cross-sectional studies of which at least one study is a high-score study. Inconclusive evidence concerns all other cases, i.e., consistent ®ndings in multiple low-score cross-sectional studies, or inconsistent ®ndings in multiple studies. Moreover, the evidence is considered to be inconclusive if only one cross-sectional study is available, irrespective of the quality of this study.

Cross-sectional studies that were rated lowest for quality according to the methodological quality list (score of 3 or less) were excluded from the analysis for the determination of the strength of evidence. A positive, a negative or no effect of a risk factor can be found in the publications reviewed. A positive effect is de®ned as an increased risk for the occurrence of neck pain due to the presence of a risk factor. In contrast, a decreased risk for the occurrence of neck pain due to the presence of a risk factor was de®ned as

182

Arie Èns et al.

TABLE I. Description of the Items in the QualityAssessment Lists Designa

Item Definition Study purpose Study design

A. B. C. D.

E. Exposure measurements

Positive if a specific, clearly stated purpose is described. Positive if the main features (description of sampling frame, distribution by age and sex) of the study population are stated. Positive if the participation rate at baseline is at least 80%. Positive if cases and controls were drawn from the same population and a clear definition of cases and controls was stated. Persons with neck pain in the last 90 days must be excluded from the control group. Positive if the response after1year of follow-up is at least 80%, or if the non-response is not selective.

F.

Positive if data on physical load at work are collected and used in the analysis. G. Positive if data on physical load at work are collected using standardized methods of acceptable quality.b H. Positive if data on psychosocial factors at work are collected and used in the analysis. I. Positive if data on psychosocial factors at work are collected using standardized methods of acceptable quality.b J. Positive if data on physical and/or psychosocial factors during leisure time are collected and used in the analysis. K. Positive if data on historical exposure at work are collected and used in the analysis. L. Positive if data on history of neck disorders, sex and age are collected and used in the analysis. M. Positive if the exposure assessment is blinded with respect to disease status. N. Positive if exposure is measured in an identical way in cases and controls. O. Positive if the exposure was assessed at a time prior to the occurrence of the outcome.

Cr Ca Pr Cr Ca Pr Cr Ca Pr Ca

Pr Cr Ca Pr Cr Ca Pr Cr Ca Pr Cr Ca Pr Cr Ca Pr Cr Ca Pr Cr Ca Pr Cr Ca Ca Ca

Outcome measurements

P. Q. R. S.

Positive if data on outcome are collected using standardized methods of acceptable quality.c Positive if incident cases are used (prospective enrolment). Positive if data on outcome are collected for at least1year. Positive if data on outcome are collected at least every 3 months.

Cr Ca Pr Ca Pr Pr

Analysis and data presentation

T.

Positive if the statistical model used is appropriate for the outcome studied and the measures of association estimated with this model are presented (including confidence intervals). Positive if the study controls for confounding. Positive if the number of cases in the multivariate analysis is at least10 times the number of independent variables in the analysis.

Cr Ca Pr

U. V.

Cr Ca Pr Cr Ca Pr

a

This column indicates whether the itemis used in the quality assessment list for cross-sectional (Cr),case^control (Ca) and/or prospective cohort studies (Pr). Thisitemisscoredpositiveifoneofthefollowingcriteriaismet:(1)fordirectmeasurements:intraclasscorrelationcoefficient > 0.60ork > 0.40;(2)forobservationalmethods:intraclasscorrelation coefficient > 0.60 or k > 0.40 for the inter- or intraobserver reliability; and (3) for self-reported data: intraclass correlation coefficient > 0.60 or k > 0.40 for the inter- or intraobserver reliability. c Thisitemis scoredpositiveifone ofthe followingcriteriaismet:(1)forself-reporteddata: intraclasscorrelation coefficient > 0.60 or k > 0.40; (2)forregistereddata: datamustshow thatregistration system is valid andreliable; and (3) for physicalexamination: intraclass correlation coefficient > 0.60 or k > 0.40 for the inter- or intraobserver reliability. b

a negative effect. No effect implied that the presence of a risk factor was neither associated with an increased nor with a decreased risk for the occurrence of neck pain. The focus of this review was on the size and direction of the risk estimate, irrespective of the level of signi®cance. A study that reported a nonsigni®cant association between a risk factor and neck pain, with no mention of the risk estimate was eliminated from the determination of the level of evidence. This ignorance of statistical signi®cance and exclusion of nonsigni®cant study results (without the men-

tion of a risk estimate), was based on the fact that in most studies no suf®cient information is presented on the possible reason for ®nding nonsigni®cant results: either there was no association or there was a lack of statistical power due to, for example, a small study population [Lang et al., 1998]. Reporting a signi®cant association without stating the risk estimate was considered as a ®nding and thus contributed to the level of evidence. Consistent ®ndings implied that the results of at least 75% of the studies investigating the effect of a certain risk


factor pointed in the same direction. The risk estimates and P-values reported by these studies should lead to the same conclusion, i.e., that a positive, negative or no effect was found in relation to neck pain. The same methods have been used in our systematic review on physical risk factors for neck pain [ArieÈns et al., 2000].

RESULTS Identi®cation of Studies and Quality Assessment Out of 1,026 studies identi®ed, 29 studies were included in this review. All studies included but one, a prospective cohort study of Viikari-Juntura et al. [1994], had a cross-sectional design. The most important reason for exclusion from this review was the use of a combined outcome measure [e.g. Bjelle et al., 1987; Veiersted and Westgaard, 1993; Westgaard et al., 1993; Engels et al., 1994; Bergqvist, 1995; Hasvold et al., 1996; HaÈgg and AstroÈm, 1997]; these studies did not report their results separately for the neck region, but combined the neck region with another body region (most often the shoulder region). The two independent reviewers agreed on inclusion or exclusion for 90% of the studies in the random sample. After discussion with a third person, consensus was reached on inclusion or exclusion of all studies. In Appendix A the results of the assessment of the methodological quality of all studies included in this review are presented. The percentage of agreement between the two independent reviewers on the methodological quality assessment was 86.3%. All disagreements between the two reviewers were discussed and resolved in a consensus meeting, and for each study, a ®nal score was given on every item. Twenty-eight studies collected and analyzed data concerning psychosocial factors at work. One additional study did not collect data on psychosocial factors at work. However, in this study, data on psychosocial factors during leisure time were collected [Westgaard and Jansen, 1992]. The items in the quality list that were most often scored negative, were those concerning the use of standardized exposure measurements of acceptable quality (items G and I) and Item L, the item that was scored positive if data on a history of neck disorders, sex, and age were collected and used in the analysis. Only once, these items were scored positive. The items on blinding of the exposure assessment and on the use of standardized methods for the outcome measures (items M and P) were scored positive only twice. Confounding was controlled for by 23 out of 29 studies (Item U) and 22 studies had also collected data on physical exposure at work. Nine cross-sectional studies scored 3 or less points on the quality list [Chang et al., 1987; Ursin et al., 1988; Flodmark and Aase, 1992; Rosecrance et al., 1992; Westgaard and Jansen, 1992; Johansson et al., 1993;

183

Johansson, 1994; Pocekay et al., 1995; IngelgaÊrd et al., 1996]. They were excluded from the determination of the level of evidence. The ®nal number of studies to be used for the level of evidence synthesis is therefore 20. Of these 20 studies, ®ve scored positive on more than 50% of the validity/precision items of the methodological quality list, and were de®ned as being high-score studies. Table II gives a brief description of the studies that have been used for the determination of the level of evidence.

Levels of Evidence As can be seen in Table II, various psychosocial factors were examined. All these factors were grouped into nine categories of psychosocial factors. The results concerning the determination of the level of evidence for these nine categories are described below and summarized in Table III. Quantitative job demands A total of 13 cross-sectional studies investigated the effect of high quantitative job demands in relation to neck pain [Kilbom et al., 1986; Musson, 1989; Linton, 1990; Kamwendo et al., 1991; Bernard et al., 1994; Hales et al., 1994; Johansson and Rubenowitz, 1994; Johansson, 1995; Ahlberg-HulteÂn et al., 1995; LagerstroÈm et al., 1995; Skov et al., 1996; Bru et al., 1996; Toomingas et al., 1997]. Three of these studies were rated as high-score studies [Kamwendo et al., 1991; Bernard et al., 1994; Hales et al., 1994]. Three studies have not been taken into account for the determination of the level of evidence because they reported a non-signi®cant relationship between psychosocial workload and neck pain without mentioning a risk estimate [Kilbom et al., 1986; Musson, 1989; Skov et al., 1996]. In their high-score study, Hales et al. [1994] reported an odds ratio of 2.4 for the relationship between increasing work pressure and self-reported neck pain. The high-score study of Bernard et al. [1994] reported an odds ratio of 1.7 for the relationship between neck pain and an increased number of hours spent working under a deadline. Kamwendo et al. [1991] showed a P-value of 0.01 for the relationship between neck pain and the following statement: `Ì have too much to do.'' Several low-score cross-sectional studies, investigating the relationship between high quantitative job demands and neck pain, con®rm the results found in the high-score studies [Linton, 1990; Johansson and Rubenowitz, 1994; LagerstroÈm et al., 1995; Johansson, 1995; Bru et al., 1996; Toomingas et al., 1997]. One low-score study [Ahlberg-HulteÂn et al., 1995] could not detect a relationship between high quantitative job demands and neck pain. Based on the results described above, it is concluded that some evidence is found for a positive relationship between high quantitative job demands and neck pain.

184

Arie Èns et al.

TABLE II. Descriptive Information ofAll Studies Included in this ReviewThat Had aTotal Quality Score of 4 or More Reference

Design MQSa Study population

Kilbom et al., 1986

Crb 5

Dartigues et al., 1988

Cr 5

Musson,1989

Cr 4

Linton,1990

Cr 6

Kamwendo et al.,1991

Cr 7

Outcome measure(s)

Female assembly line workers of two electronic manufacturing companies Response at baseline 77% (n 106) Aworking population (n 990)

Severity of self-reported neck symptoms

Work-related factors Overtimework (ns);c Perceivedpsychological stress at work (ns); Work satisfaction (ns); Number of breaks and rest pauses at work (ns)

Self-reported recurrent cervical pain syndrome

Workers using various types of impact tools (n 445) Response at baseline 38% (n 169) Full-time employees daytime working (n 22,180)

Self-reported regularly pain or stiffness in the neck

Work-related risk factors Conflict related to work (OR 3.1, 2.0^4.8);d Non-work-related risk factors Conflict related to family (OR 1.8,1.1^3.0); Work-related factors Time pressure (ns)

Female medical secretaries and office personnel

Self-reported neck pain


Response at baseline 96% (n 420)

MÌkelÌ et al., 1991

Cr 9

Ignatius et al., 1993

Cr 6

Bernard et al., 1994

Cr 9

Finnish adults drawn from the population register, representing the Finnish adult population of 30 years and older Response at baseline 90% (n 7,217) Female typists working in the Government Housing Department Response at baseline 52% (n 170) Newspaper employees using video display terminals

Chronic neck syndrome

Telecommunication employees using video display terminals for at least 6 h per day Response at baseline 96% (n 512)

Work-related risk factors Monotonous work (OR 2.25^2.95);e Overall psychosocial score (OR 1.89^2.57); Poor work content (OR 1.94^2.47); Low social support (OR 1.38^2.57); High psychosocial work load (OR 1.24^1.49) Work-related risk factors Poorly experienced psychosocial work environment (P 0.004);f Interesting and stimulating work (ns); Work variation (ns); Friendly spirit of cooperation with fellow workers (P 0.013); Help and support if you run into difficulties in your work (ns); Ability to influence working conditions (P 0.001); Too much to do (P 0.010); Good contact and cooperation with superiors (ns); Demands of your work too great (ns); Anxiety feelings about possible reorganization or new techniques in your work (ns) Work-related factors (age 30^64 years) Mental stress at work (OR 1.20,1.12^1.28) Work-related risk factors (age > 64 years) Mental stress at work (OR 1.27,1.11^1.46)


Work-related factors No rest other than lunch breaks (ns)

Self-reported neck symptoms

Work-related risk factors Number of hours spent under a deadline per week (OR 1.7,1.4^3.0); Work variance (OR 1.7,1.2^2.5); Number of breaks (ns); Job control (ns); Job security (ns); Interaction with coworkers or customers (ns); Group conflict (ns) Non-work-related risk factors Lack of social support from spouses and friends (ns) Work-related risk factors Routine work lacking decision making opportunities (OR 4.2, 2.1^8.6), Lack of productivity standard (OR 3.5,1.5^8.3); Fear of being replaced by computers (OR 3.0,1.5^6.1); High information processing demands (OR 3.0,1.4^6.2); Job requires a variety of tasks (OR 2.9,1.5^5.8); Increasing work pressure (OR 2.4,1.1^5.5)


Hales et al.,1994 Cr 7

Psychosocial risk factor(s) and strength of association

Self-reported neck disorders


185

TABLE II. (Continued) Reference


Johansson and Rubenowitz, 1994

Cr 5

Blue and white collar workers from 8 large metal industry companies Response at baseline 90% (n 450)

Outcome measure(s) Self-reported neck symptoms

Self-reported neck symptoms, symptoms must be work-related

Viikari-Juntura et al.,1994

Ahlberg-Hulten et al.,1995

Prg 9

Cr 5

Male machine operators, carpenters and office workers

Self-reported neck pain, change from1984 to1987: * none to moderate

Response at baseline 69% (n 2,222) Response at follow-up 82%) (n 1,832) Female nurses and nurse's aides

*

none to severe

*

persistent severe

Self-reported pain in the neck

Participation rate 79% (n 90)

Bergqvist et al., 1995

Cr 6

Johansson, 1995

Cr 6

Office workers (n 353)

Tension neck syndrome

Response Qj 92% Response PEk 91% Response WAl 82% Home care workers (n 305)


Self-reported work related neck symptoms


Work-related factors (blue collar workers) Low influence on and control over work (ns); Poor supervisor climate (P < 0:05); Low stimulus from the work (ns); Poor relations with fellow workers (ns); High psychological work load (P < 0:001) Work-related factors (white collar workers) Low influence on and control over work (ns); Poor supervisor climate (ns); Low stimulus from the work (ns); Poor relations with fellow workers (ns); High psychological work load (ns) Work-related risk factors (blue collar workers) Low influence on and control over work (ns); Poor supervisor climate (P < 0:05); Low stimulus from the work (P < 0:05; Poor relations with fellow workers (ns); High psychological work load (P < 0:001); Work-related risk factors (white collar workers) Low influence on and control over work (P < 0:05); Poor supervisor climate (ns); Low stimulus from the work (ns); Poor relations with fellow workers (ns); High psychological work load (P < 0:01) Work-related factors Job satisfaction (ns) Work-related factors Job satisfaction (OR 1.7,1.1^2.6) Work-related factors Job satisfaction (ns) Work-related factors Feeling of isolation (0.01, P 0.92);h Poor relations with superiors (ÿ0.14, P 0.40); Conflicts (0.11, P 0.52); Stress (0.08, P 0.72); Intensity of authority over decisions (0.05, P 0.71); High psychological demands (0.00, P 0.97); Lowskill utilization (ÿ0.03, P 0.73); High job strain (ÿ0.43, P 0.67; 0.59, P 0.62)i Work-related factors Limited rest break opportunities (OR 7.4, 3.1^17.4)

Work-related factors Low influence and control over work (RR 1.27,1.00^1.62);m Poor supervisor climate (RR 1.23, 0.99^1.53); Low stimulus from the work itself (RR 1.33,1.05^1.67); Poor relationships with fellow workers (RR 1.19, 0.94^1.50); High psychological work load (RR 1.52,1.20^1.94; P < 0.001) Work-related factors Low influence and control over work (RR 1.30, 0.93^1.81); Poor supervisor climate (RR 1.29, 0.93^1.79); Low stimulus from the work itself (RR 1.52,1.10^2.11); Poor relationships with fellow workers (RR 1.20; 0.87^1.65); High psychological work load (RR 1.83,1.28^2.61; P < 0.001)

186

Arie Èns et al.

TABLE II. (Continued) Reference


LagerstrÎm et al., Cr 1995 5

Female nursing personnel of a hospital

Outcome measure(s) Self-reported ongoing neck symptoms

Work-related risk factors Low work commitment (OR 1.67,1.10^2.60; OR 1.65, 1.07^2.54); Low support from superiors (OR 2.08, 1.32^3.26; OR 2.03,1.28^3.16); High work demand (ns); Lack of stimulation (ns); Low work control (ns)

Self-reported severe ongoing neck symptoms

Work-related risk factors Low work commitment (ns); Low support from superiors (ns); High work demand (OR 1.82,1.14^2.92; OR 1.82, 1.14^2.92); Lack of stimulation (ns); Low work control (ns) Work-related factors High workload variability (OR 1.2,1.0^1.4)


Tharr,1995

Cr 6

Bru et al., 1996

Cr 5

Skov et al., 1996

Cr 6

Teleservice representatives drawn from 2 teleservice centers Response at baseline 95% (n 108) Female hospital staff Response at baseline 85% (n 586) Random 8% sample of the members of the association of Danish Active Salespeople


Neck pain index (based on self-reported data)

Work-related factors Work overload (P 0.004); Poor social relations (P 0.005); Poor work content (P 0.03)


Work-related factors High demands in thework (ns); Variation in the work (highest quartile is reference value), next to highest quartile (OR 1.78,1.16^2.73), next to lowest quartile (ns), lowest quartile (OR 1.82,1.23^2.69); Control over time, low compared to high control (OR 1.44, 1.07^1.93), medium compared to high control (ns); Perceived competition, high compared to low competition (OR 1.44,1.08^1.91), medium compared to low perceived competition (ns) Work-related factors High psychological demands (PR 1.5,1.1^2.0);n High decision latitude (ns); High social support (PR 1.6,1.1^2.3); High job strain (PR 1.6,1.1^2.2) Work-related factors High psychological demands (PR 2.0,1.1^3.7); High decision latitude (ns); High social support (ns); High job strain (PR 2.1,1.2^3.7) Work-related factors High decision latitude (ns); High job strain (ns) Work-related factors High psychological demands (ns); High decision latitude (ns); High social support (PR 2.7,1.1^6.7); High job strain (ns) Work-related factors Good relation with workmates/foreman (P < 0:01; P < 0:01), Low work satisfaction (P < 0:04), Stress at work (P < 0:001) Work-related factors Good relation with workmates/foreman (ns),Work satisfaction (ns), Stress at work (P < 0:005)

Response at baseline 66% (n 1,306) Toomingas et al., Cr 1997 4

Male furniture movers, female medical secretaries and males and females of the working population


Zettenberg et al., Cr 1997 6

Car assembly workers (n 564)

Self-reported neck symptoms in past12 months

Neck sign: neck tenderness

Neck sign: neck movement restriction Neck Syndrome: neck tension syndrome Self-reported neck complaints

Neck myalgia

a


Methodologicalquality score; bCross-sectionalstudy; cNot significant; dOdds ratio and 95% confidence interval; eSeveralodds ratios are presented within this range fordifferent age groups; fP-value; Prospective cohort study; hCoefficient and corresponding P-value; iIf two analyses are carried out,results of both analyses are presented; jQuestionnaire; kPhysicalexamination; lWorkplace assessment; mRate ratio and 95% confidence interval; nPrevalence ratio and 95% confidence interval.

g


187

TABLE III. Summary of the Results Concerning the Level of Evidence Synthesis Direction of the association

Number of high and low-score studiesa

Number of studies with positive effectb

Quantitative job demands

High

Social support at work

Low

Supervisor support

Low

Coworker support

Low

Conflicts at work

Yes

Conflicts in leisure time

Yes

Job control

Low

Skill discretion

Low

High: 3 Low: 7 High: 1 Low: 8 High: 0 Low: 4 High: 1 Low: 1 High: 0 Low: 2 High: 0 Low: 1 High: 2 Low: 4 High: 1 Low: 6 High: 1 Low: 1 High: 0 Low: 3 High: 1 Low: 3 High: 1 Low: 0 High: 0 Low: 1

High: 3 Low: 6 High: 1 Low:7 High: 0 Low: 3 High: 1 Low: 1 High: 0 Low: 1 High: 0 Low: 1 High: 2 Low: 3 High: 1 Low: 5 High: 1 Low: 1 High: 0 Low: 2 High: 1 Low: 3 High: 1 Low: 0 High: 0 Low: 1

Risk factor

High Job strain

High

Job satisfaction

Low

Job security

Low

Rest break opportunities

Limited

a

Level of evidence Some Some Inconclusive Some Inconclusive Inconclusive Some Some Some Inconclusive Some Inconclusive Inconclusive

Number of high-score (high) and low-score (low) studies that are used for the determination of the level of evidence. Number of high-score (high) andlow-score (low) studies that reported a positive effect of a risk factor,i.e., anincreased risk for the occurrence of neckpain.

b

Social support Ten cross-sectional studies investigated the relationship between work-related social support and neck pain [Linton, 1990; Kamwendo et al., 1991; Bernard et al., 1994; Johansson and Rubenowitz, 1994; Johansson, 1995; LagerstroÈm et al., 1995; Ahlberg-HulteÂn et al., 1995; Bru et al., 1996; Toomingas et al., 1997; Zettenberg et al., 1997]. Two of these studies were classi®ed as being high-score studies [Kamwendo et al., 1991; Bernard et al., 1994]. One of these high-score studies was not taken into account for the determination of the level of evidence, since this study only reported that the relationship between low social support and neck pain was not signi®cant, without mentioning a risk estimate [Bernard et al., 1994]. Kamwendo et al. [1991] reported a P-value of 0.013 for the relationship between neck pain and `à poor spirit and cooperation with fellow workers'' in their high-score study. Seven of eight low-score cross-sectional studies [Linton, 1990; Johansson and

Rubenowitz, 1994; Johansson, 1995; LagerstroÈm et al., 1995; Bru et al., 1996; Toomingas et al., 1997; Zettenberg et al., 1997] con®rm the results of Kamwendo et al. [1991], leading to the conclusion that there is some evidence of a positive relationship between poor social support at work and neck pain. With respect to work-related social support, supervisor support and coworker support can be distinguished. The relationship between supervisor support and neck pain was investigated by ®ve cross-sectional studies [Kamwendo et al., 1991; Johansson and Rubenowitz, 1994; AhlbergHulteÂn et al., 1995; Johansson, 1995; LagerstroÈm et al., 1995], one of which was a high-score study [Kamwendo et al., 1991]. Despite the high score of Kamwendo et al. [1991], a non-signi®cant relationship was reported, without the mention of a risk estimate. Three of the remaining four low-score cross-sectional studies suggested a positive relationship between poor supervisor support and neck pain. However, due to the low-score of these cross-sectional

188

Arie Èns et al.

studies, it is concluded that there is inconclusive evidence for the relationship between low supervisor support and neck pain. Results regarding the relationship between coworker support and neck pain are reported by three cross-sectional studies [Kamwendo et al., 1991; Johansson and Rubenowitz, 1994; Johansson, 1995]. One study was not included in the determination of the level of evidence, since only a nonsigni®cant relationship without a risk estimate was mentioned [Johansson and Rubenowitz, 1994]. In their highscore study, Kamwendo et al. [1991] reported a P-value of 0.013 for the relationship between poor coworker support and neck pain. In the low-score study of Johansson [1995] this result was con®rmed, leading to the conclusion that there is some evidence for a positive relationship between poor coworker support and neck pain. Finally, as the only study in this review, the high score study of Bernard et al. [1994] studied the relationship between neck pain and the lack of social support by friends and family. However, they reported a non-signi®cant association and no risk estimate was mentioned. This leads to the conclusion that there is inconclusive evidence for such a relationship.

between job control and neck pain was not signi®cant without the report of a risk estimate [Bernard et al., 1994; LagerstroÈm et al., 1995; Toomingas et al., 1997]. Of the remaining six studies, two were quali®ed as high-score studies [Kamwendo et al., 1991; Hales et al., 1994]. Both these studies reported results suggesting a positive relationship between low job control and neck pain. First, Hales et al. [1994] reported an odds ratio of 4.2 for the relationship between neck pain and routine work lacking decision making opportunities. Second, Kamwendo et al. [1994] showed a P-value of 0.001 for the relationship between low ability to in¯uence working conditions and neck pain. Three lowscore cross-sectional studies con®rmed the results of the high-score studies [Johansson and Rubenowitz, 1994; Johansson, 1995; Skov et al., 1996]. One low-score crosssectional study could not identify a relationship between job control and neck pain [Ahlberg-HulteÂn et al., 1995]. Based on the results of six cross-sectional studies, of which two were of high-score studies, it is concluded that there is some evidence for a positive relationship between low job control and neck pain.

Con¯icts

Nine cross-sectional studies reported results on the relationship between low skill discretion and neck pain. Two studies were high-score studies [Kamwendo et al., 1991; Bernard et al., 1994], seven studies were classi®ed as lowscore studies [Linton, 1990; Johansson and Rubenowitz, 1994; Ahlberg-HulteÂn et al., 1995; Johansson, 1995; LagerstroÈm et al., 1995; Skov et al., 1996; Bru et al., 1996]. One low-score and one high-score study reported that the relationship between skill discretion and neck pain was not signi®cant without reporting a risk estimate [Kamwendo et al., 1991; LagerstroÈm et al., 1995]. In their high-score study, Bernard et al. [1994] reported an odds ratio of 1.7 for the relationship between neck pain and low work variance. Five low-score cross-sectional studies con®rmed the results of Bernard et al. [1994], all suggesting a positive relationship between low skill discretion and neck pain [Linton, 1990; Johansson and Rubenowitz, 1994; Johansson, 1995; Bru et al., 1996; Skov et al., 1996]. One low-score crosssectional study could not identify a relationship between skill discretion and neck pain [Ahlberg-HulteÂn et al., 1995]. Based on the results of the six studies that were used for the determination of the level of evidence, it is concluded that there is some evidence for a positive relationship between low skill discretion and neck pain. On the contrary, two studies investigated the effect of high skill discretion on neck pain [Hales et al., 1994; Tharr, 1995]. In their high-score study, Hales et al. [1994] reported an odds ratio of 2.9 for the relationship between neck pain and a variety of job tasks. Tharr [1995] reported an odds ratio of 1.2 for high work load variability in relation to neck

Three cross-sectional studies investigated the relationship between con¯icts at work and neck pain [Dartiques et al., 1988; Bernard et al., 1994; Ahlberg-HulteÂn et al., 1995]. The study of Bernard et al. [1994], a high-score study, however, reported that the relationship between neck pain and con¯icts at work was not signi®cant, without reporting a risk estimate. The other two studies [AhlbergHulteÂn et al., 1995; Dartiques et al., 1988] were low-score cross-sectional studies. Due to the low score of these two studies, it is concluded that there is inconclusive evidence for a relationship between con¯icts at work and neck pain. One study also investigated the effect of non-workrelated con¯icts on the occurrence of neck pain [Dartiques et al., 1988]. An odds ratio of 1.8 was reported for this relationship. Due to the fact that no other studies in this review investigated this relationship, inconclusive evidence was found for the relationship between non-work-related con¯icts and neck pain. Job control Nine cross-sectional studies studied the risk factor job control in relation to neck pain [Kamwendo et al., 1991; Bernard et al., 1994; Hales et al., 1994; Johansson and Rubenowitz, 1994; LagerstroÈm et al., 1995; Johansson, 1995; Ahlberg-HulteÂn et al., 1995; Skov et al., 1996; Toomingas et al., 1997]. Three of these studies were not taken into account because they stated that the relationship

Skill discretion


pain. Based on the ®ndings of these two cross-sectional studies, of which one is a high-score study, it is concluded that there is some evidence for a positive effect of high skill discretion in relation to neck pain. Job strain A total of four cross-sectional studies investigated the relationship between high job strain and neck pain [Kilbom et al., 1986; Ahlberg-HulteÂn et al., 1995; Toomingas et al., 1997; Zettenberg et al., 1997]. One of these studies reported a non-signi®cant relationship between perceived psychological stress at work and neck pain, but did not mention a risk estimate [Kilbom et al., 1986]. Three studies remained for the determination of the level of evidence. Due to the low score of these studies, and irrespective of their results, it is concluded that there is inconclusive evidence for a relationship between high job strain and neck pain. Job satisfaction Three low-score cross-sectional studies [Kilbom et al., 1986; Bru et al., 1996; Zettenberg et al., 1997] and one highscore prospective study [Viikari-Juntura et al., 1994] investigated the relationship between job satisfaction and neck pain. Viikari-Juntura et al. [1994] reported an odds ratio of 1.7 for the relationship between low job satisfaction and change in neck pain from no neck pain at baseline till severe neck pain at follow-up. This result suggested a positive relationship between low job satisfaction and the development of neck pain. Zettenberg et al. [1997] and Bru et al. [1996] reported P-values of 0.04 and 0.03, respectively, for the relationship between low job satisfaction and neck pain. Kilbom et al. [1986] reported that this relationship was not signi®cant, without mentioning a risk estimate. Based on the results of one high-score prospective study of ViikariJuntura et al. [1994] and two low-score cross-sectional studies of Zettenberg et al. [1997] and Bru et al. [1996], it is concluded that there is some evidence for a positive relationship between low job satisfaction and neck pain. One additional study investigated the relationship between neck pain and low work commitment [LagerstroÈm et al., 1995]. If ``low work commitment'' is considered to represent low job satisfaction, the study of LagerstroÈm et al. [1995] should also be taken into account for the determination of the level of evidence. The conclusion, that there is some evidence for a positive relationship between low job satisfaction and neck pain, will not change if this additional study is added. Job security A total of three cross-sectional studies, all de®ned as high-score studies, reported results concerning the relation-

189

ship between low job security and neck pain [Kamwendo et al., 1991; Bernard et al., 1994; Hales et al., 1994]. Two of these studies reported a non-signi®cant relationship between job security and neck pain, without mentioning a risk estimate. Hales et al. [1994] reported an odds ratio of 3.0 for the relationship between ``the fear of being replaced by a computer'' and neck pain in their high-score study. Since only one study reported results to determine the level of evidence, it is concluded that there is inconclusive evidence for the relationship between low job security and neck pain. Rest break opportunities Of the four cross-sectional studies that reported results on the relationship between neck pain and rest break opportunities, three studies stated that this relationship was not signi®cant. No risk estimate was presented [Kilbom et al., 1986; Bernard et al., 1994; Bergqvist et al., 1995]. Bergqvist et al. [1995] reported an odds ratio of 7.4 for the relationship between ``limited rest break opportunities'' and neck pain. Based on only one low-score cross-sectional study, it is concluded that there is inconclusive evidence for the relationship between rest break opportunities and neck pain.

DISCUSSION In order to identify psychosocial risk factors for neck pain, a systematic review of the literature was carried out. The results showed some evidence for a positive relationship between neck pain and high quantitative job demands, poor social (coworker) support, low job control, low skill discretion, and low job satisfaction. Inconclusive evidence was found for the relationship between neck pain and poor supervisor support, con¯icts at work, low job security, high job strain, and limited rest break opportunities (Table III). Other factors, such as feelings of isolation and lack of productivity standards were taken into consideration by some of the studies in this review, although these factors were not used in the determination of the level of evidence. The reason for this was that it was dif®cult to place these factors within any of the nine categories of risk factors identi®ed in this review. Furthermore, in three studies very general measures for psychosocial exposure at work were used, such as ``mental stress at work'' or `òverall psychosocial score'' [Linton, 1990; Kamwendo et al., 1991; MaÈkelaÈ et al., 1991]. Since the focus of this review was on speci®c aspects of the psychosocial exposure, these general measures were not discussed. The risk factor ``skill discretion'' was studied in different ways. Eight studies investigated the effect of low skill discretion on the occurrence of neck pain, while two studies looked at the effect of high skill discretion in relation to neck pain. For both high and low skill discretion it was

190

Arie Èns et al.

concluded that there is some evidence of a relationship with neck pain, suggesting a U-shaped relationship between skill discretion and neck pain. Several investigators suggest three mechanisms that account for possible associations between psychosocial factors and musculoskeletal disorders [Bongers et al., 1993; Hales and Bernard, 1996; Sauter and Swanson, 1996]. Firstly, they suggest that psychosocial demands can exceed an individual's coping capabilities, resulting in a stress response, which, in turn, can produce muscle tension or static loading of the muscles or generate other physiological responses that may result in neck pain. As a second mechanism, they suggest that psychosocial demands may affect the awareness and reporting of musculoskeletal disorders, or increase its attribution to the work environment. As a third possible mechanism, it is stated that, in a certain situation, psychosocial demands may be highly correlated with physical demands. This suggests that any association between a psychosocial risk factor and musculoskeletal disorders may actually re¯ect a relationship between a physical risk factor and musculoskeletal disorders. As stated above it is hypothesized that psychosocial and physical demands may be highly correlated. Studies on psychosocial risk factors for neck pain should therefore also take the physical workload into account in their analyses. In this review, several studies did not assess the physical workload [Ursin et al., 1988; Flodmark and Aase, 1992; Westgaard and Jansen, 1992; Ahlberg-HulteÂn et al., 1995; LagerstroÈm et al., 1995; Toomingas et al., 1997; Zettenberg et al., 1997]. In order to test the effect of these studies on the level of evidence of the psychosocial factors, a sensitivity analysis was carried out. In this analysis, studies that did not assess or control for work-related physical exposure (studies that scored negative or unknown on Item F of the quality list) were eliminated. Without these studies, the level of evidence for the different psychosocial variables in this review was determined again. Exclusion of these studies had no effect on the results regarding the level of evidence for the different psychosocial factors in this review. Three other literature reviews were found focusing on psychosocial risk factors for neck pain. Bongers et al. [1993] discussed the literature for psychosocial risk factors at work for musculoskeletal disorders. They concluded that a relationship between psychosocial variables, such as monotonous work, time pressure, poor work content, and high workload, and symptoms of the neck or shoulders seemed likely. For the risk factor social support, they found contradictory results, whereas in our review some evidence was found for coworker support and inconclusive evidence was found for supervisor support. However, a comparison between the results of the review of Bongers et al. [1993] and this review is dif®cult, since Bongers et al. [1993] also included studies that combined neck symptoms with shoulder symptoms. Hales and Bernard [1996] have

critically examined the literature to describe psychosocial risk factors that are associated with neck disorders. Hales and Bernard stated that high work load, perceived time pressure, work pressure, high work load variability, poor work content, and monotonous work are associated with musculoskeletal complaints of the upper extremities. However, they stated that most of the studies based the case de®nition of musculoskeletal disorders on self-reports of neck and shoulder symptoms. In the present review, studies using a combined outcome measure (e.g., the combination of the neck region and shoulder region) were not included, therefore making it hard to compare the results of this review with the results found by Hales and Bernard [1996]. In the NIOSH review of Bernard [1997] neck disorders were combined with shoulder, elbow, hand, and wrist disorders. Bernard [1997] concluded that intensi®ed workload, monotonous work, and low levels of support have a positive association with these upper extremity disorders. Moreover, lack of control over the job and low job satisfaction were also positively associated with these disorders, although not as strongly. Bernard [1997] also stated that the evidence of the relationships between these factors and upper extremity disorders was stronger for disorders related to the neck/shoulder region in comparison to the hand/wrist region. Again, the comparison of the results of the review of Bernard with this review is dif®cult, since no results for the neck as a separate region were reported by Bernard. For this review several databases were systematically searched to identify all relevant studies. It is crucial to ®nd all possible studies, involving the subject of this review. Many risk factor studies do not consider one outcome measure, but investigate several outcome measures, neck pain being one of them. If, in these studies, the main focus is not on neck pain but, for example, on low back pain, these studies might have used keywords relating only to low back pain instead of also to neck pain. Consequently, it is possible that these studies were missed during the literature search. The number of studies that was found in the literature for several psychosocial factors in this review was small. One additional study could have changed the conclusion regarding the level of evidence. Consequently, missing a study, even if this was not selective with regard to the study results, may have in¯uenced the conclusion regarding the level of evidence for these psychosocial factors. The most important reason for exclusion of a study from this review was the fact that results of a study were not reported for the neck region separately. A lot of studies do not use neck pain as an outcome measure, but used a combination of neck and/or shoulders pain as the outcome measure [e.g., Bjelle et al., 1987; Veiersted and Westgaard, 1993; Westgaard et al., 1993; Engels et al, 1994; Bergqvist, 1995; Hasvold et al., 1996; Hagg and Astrom, 1997]. Since the objective of this review was to identify risk factors for neck pain, these studies were excluded. In the excluded


studies, it is often unclear what was meant by neck and/or shoulder pain. Pain in the proximal part of the upper arm may also have been included in these studies. Other risk factors may be of in¯uence to determine whether pain in this region will exist, and therefore these studies were excluded. However, on the other hand this may have led to the exclusion of studies that did actually investigate the neck region. Most of the studies identi®ed were cross-sectional studies. No case±control study and only one prospective cohort study entered this review [Viikari-Juntura et al., 1994]. In cross-sectional research both risk factors and outcome are measured at the same time. Therefore, in crosssectional research, cause and effect cannot be distinguished and a causal relationship can hardly be established. The reason to include cross-sectional studies in this review, despite of this disadvantage, was that most research on risk factors for neck pain is actually carried out with the use of a cross-sectional design. Although perhaps desirable from a purely methodological standpoint, it would not be acceptable to neglect this large amount of information obtained from cross-sectional research. The maximum level of evidence that consequently could be reached was ``some evidence'', due to the fact that there was only one (high score) prospective study included in this review. A quality list was constructed to assess the methodological quality of the studies in this review. This list consists of several items in different categories concerning information, validity, and precision. A total quality score was calculated by counting the number of validity and precision items in the criteria list that were scored positively. Based on this obtained total quality score, studies were labeled as either being a high-score study or a low-score study. Four levels of evidence were de®ned to establish the strength of evidence of a relationship between a risk factor and neck pain. Obviously, this procedure and our rating system had a considerable in¯uence on the assessment of the level of evidence, meaning that changes in this procedure may have had an impact on the results. The methodology of rating of the methodological quality of studies is widely used in systematic reviews of the effectiveness of certain clinical treatments, but is new and still in an experimental stage for systematic reviews of observational studies. No established guidelines for rating procedures for such studies are avail-

191

able yet. In the quality list developed for this review some items, especially the items on the use of standardized methods for the collection of exposure and outcome (items G,I and P), in retrospect did not really discriminate between high- and low-score studies, since almost all studies scored negative on these items. If these three items were not taken into account the number of high-quality studies would increase from 5 to 12 [Linton, 1990; Kamwendo et al., 1991; MaÈkelaÈ et al.,1991; Ignatius et al., 1993; Bernard et al., 1994; Hales et al., 1994; Viikari-Juntura et al., 1994; Bergqvist et al., 1995; Johansson, 1995; Tharr, 1995; Skov et al., 1996; Zettenberg et al., 1997]. This sensitivity analysis would lead to the conclusion that, in addition to the previously mentioned factors, some evidence is also found for low supervisor support (data not shown). Many studies in this review just reported that the relationship between a risk factor and neck pain was not signi®cant, without mentioning the risk estimate. Since the direction of such a result is unclear, it was decided not to take these studies into account for the determination of the level of evidence. If the report of a non-signi®cant relationship were interpreted as no relationship, and these results were taken into account for the determination of the levels of evidence, some evidence would be found for a positive relationship between neck pain and low social support, low job satisfaction, and high skill discretion. In addition some evidence would be found of no relationship between limited rest break opportunities and neck pain. Inconclusive evidence would be found for all other risk factors discussed in this review. In conclusion, this systematic review shows some evidence of a positive relationship between neck pain and the following psychosocial risk factors: high quantitative job demands, low social (coworker) support, low job control, high as well as low skill discretion, and low job satisfaction. Furthermore, it should be concluded that, due to the study design, study population, and data analysis of observational studies, it appeared to be dif®cult to construct a valid and reliable quality assessment list that could be used to determine the quality of observational studies. Although major pitfalls still have to be accounted for, we still feel that there is much to gain from a systematic and transparent method for the review of observational studies.

APPENDIX A: METHODOLOGICAL QUALITY SCORES The scores on the items of quality assessment list for all studies in the review.The letters on the first row correspond with the letters in front of the item definitions inTable I. Reference

A

B

C

E

F

G

H

I

J

K

L

M

P

Kilbom et al.,1986 Chang et al.,1987

ÿ ÿ

b

? ?

? ?

ÿ

ÿ ÿ

ÿ ÿ

? ?

R

S

T

U

V Totala

ÿ ? ÿ ÿ ÿ

5 3

192

Arie Èns et al.

Reference

A

B

C

Dartiques et al.,1988 Ursin et al.,1988 Musson,1989 Linton,1990 Kamwendo et al.,1991 MÌkelÌ et al.,1991 Flodmark and Aase,1992 Rosecrance et al.,1992 Westgaard and Jansen,1992 Ignatius et al.,1993 Johansson et al.,1993 Bernard et al.,1994 Hales et al.,1994 Johansson and Rubenowitz, 1994 Johansson,1994 Viikari-Juntura et al.,1994 Ahlberg-HultÅn et al.,1995 Bergqvist et al.,1995 Johansson,1995 LagerstrÎm et al.,1995 Pockacy et al.,1995 Tharr,1995 Bru et al.,1996 IngelgÔrd et al.,1996 Skov et al.,1996 Toomingas et al.,1997 Zettenberg et al.,1997

ÿ ÿ ? ÿ ÿ

? ÿ ? ÿ ? ÿ ?

?

? ÿ ÿ ÿ ÿ ÿ ÿ ÿ

a b

E

F

G

H

I

J

K

L

M

P

ÿ ÿ ÿ

? ÿ ? ? ? ÿ ? ÿ ? ? ? ? ?

ÿ

? ? ? ? ? ? ? ÿ ? ? ? ? ?

ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ

ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ

? ÿ ÿ ÿ ÿ ? ÿ ÿ ÿ ÿ ÿ ÿ ÿ

ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ

? ? ? ? ? ? ? ? ? ? ? ? ?

ÿ ? ? ÿ

? ? ÿ ? ? ? ? ? ? ? ? ? ÿ

? ? ? ? ? ? ? ? ? ? ? ? ?

ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ

ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ

ÿ ? ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ

ÿ

? ? ? ? ? ? ? ? ? ? ? ?

ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ ÿ

R

S

ÿ

T

U

V Totala

ÿ ÿ ÿ ÿ ÿ ÿ

ÿ ÿ ÿ

? ? ÿ ÿ ÿ ? ÿ ? ÿ

5 3 4 6 7* 9* 2 2 3 6 3 7* 7* 5

ÿ ÿ ÿ ÿ ÿ

ÿ ÿ

ÿ ? ? ? ÿ

3 9* 5 6 6 5 3 6 5 2 6 4 6

Totalscore calculated by counting the number of positive validity/precision items.Studies marked with an asterisk (*) are high-score studies. An empty cellin the table implies that the item wasnot in the methodologicalquality list for the study design of the study at issue.

REFERENCES Ahlberg-HulteÂn GK, Theorell T, Sigala F. 1995. Social support, job strain and musculoskeletal pain among female health care personnel. Scand J Work Environ Health 21:435±439. ArieÈns GAM, van Mechelen W, Bongers PM, Bouter LM, van der Wal G. 2000. Physical risk factors for neck pain. Scand J Work Environ Health 26:7±19. Bergqvist U, Wolgast E, Nilsson B, Voss M. 1995. The in¯uence of VDT work on musculoskeletal disorders. Ergonomics 38:754±762. Bergqvist U. 1995. Visual Display terminal work Ð a perspective on long-term changes and discomforts. Int J Ind Ergon 16:201±209. Bernard BP, editor. 1997. Musculoskeletal disorders and workplace factors: a critical review of epidemiologic evidence for work-related musculoskeletal disorders of the neck, upper extremity, and lower back. Cincinnati, OH. U.S. Department of Health and Human Services. Bernard B, Sauter S, Fine L, Petersen M, Hales T. 1994. Job task and psychosocial risk factors for work-related musculoskeletal disorders among newspaper employees. Scand J Work Environ Health 20:417± 426. Bjelle A, Hagberg M, Michaelson G. 1987. Work-related shoulderneck complaints in industry: a pilot study. Br J Rheum 26:365±369.

Bongers PM, de Winter CR, Kompier MAJ, Hildebrandt VH. 1993. Psychosocial factors at work and musculoskeletal disease. Scand J Work Environ Health 19:297±312. Borghouts JAJ, Koes BW, Bouter LM. 1998. The clinical course and prognostic factors of non-speci®c neck pain: a systematic review. Pain 77:1±13. Bru E, Mykletun RJ, Svebak S. 1996. Work-related stress and musculoskeletal pain among female hospital staff. Work Stress 10: 309±321. Chang WS, Bejjani FJ, Chyan D, Bellegarde M. 1987. Occupational musculoskeletal disorders of visual artists. A questionnaire and video analysis. Ergonomics 30:33±46. Dartiques JF, Henry P, Puymirat E, Commenges D, Peytour P, Gagnon M. 1988. Prevalence and risk factors of recurrent cervical pain syndrome in a working population. Neuroepidem 7:99±105. Engels JA, van der Gulden JWJ, Senden TF, Hertog CAWM, Kolk JJ, Binkhorst RA. 1994. Physical work load and its assessment among the nursing staff in nursing homes. J Occup Med 36:338±345. Flodmark BT, Aase G. 1992. Musculoskeletal symptoms and type A behaviour in blue collar workers. Br J Ind Med 49:683±687. Ê stroÈm A. 1997. Load pattern and pressure pain threshold HaÈgg GM, A in the upper trapezius muscle and psychosocial factors in medical


secretaries with and without shoulder/neck disorders. Int Arch Occup Environ Health 69:423±432. Hales TR, Bernard BP. Epidemiology of workrelated musculoskeletal disorders. 1996. Orth Clin North Am 27:679±709. Hales TR, Sauter SL, Peterson MR, Fine LJ, Putz-Anderson V, Schleifer LR, Ochs TT, Bernard BP. 1994. Musculoskeletal disorders among visual display terminal users in a telecommunication company. Ergonomics 37:1603±1621. Hasvold T, Johnsen R, Fùrde OH. 1996. Non-migrainous headache, neck or shoulder pain, and migraine Ð differences in association with background factors in a city population. Scand J Prim Health Care 14:92±99.

193

Musson Y, Burdorf A, van Drimmelen D. 1989. Exposure to shock and vibration and symptoms in workers using impact power tools. Ann Occup Hyg 33:85±96. Pocekay D, McCurdy SA, Samuels SJ, Hammond SK, Schenker MB. 1995. A cross-sectional study of musculoskeletal symptoms and risk factors in semiconductor workers. Am J Ind Med 28: 861±871. Rosecrance JC, Cook TM, Wadsworth CT. 1992. Prevalence of musculoskeletal disorders and related job factors in 900 newspaper workers. Adv Ind Ergon Safety 4:141±146.

È rtengren R. 1996. Psychosocial IngelgaÊrd A, Karlsson H, NonaÊs K, O and physical work environment factors at three workplaces dealing with materials handling. Int J Ind Ergon 17:209±220.

Sauter SL, Swanson NG. 1996. An ecological model of musculoskeletal disorders in of®ce work. In: SD Moon SD and Suater SL, editors. Beyond biomechanics: psychosocial aspects of musculoskeletal disorders in of®ce work. London: Taylor & Francis, pp 3±21.

Ignatius YTS, Yee YT, Yan LT. 1993. Self-reported musculoskeletal problems amongst typists and possible risk factors. J Hum Ergol 22:83±93.

Skov T, Borg V, érhede E. 1996. Psychosocial and physical risk factors for musculoskeletal disorders of the neck, shoulders, and lower back in salespeople. Occup Environ Med 53:351±356.

Ê . 1994. Psychosocial and physical working conditions Johansson JA and associated musculoskeletal symptoms among operators in ®ve plants using arc welding in robot stations. Int J Hum Factors Manu 4:191±204.

Stock SR. 1991. Workplace ergonomic factors and the development of musculoskeletal disorders of the neck and upper limbs: a metaanalysis. Am J Ind Med 19:87±107.

Ê . 1995. Psychosocial work factors, physical work load Johansson JA and associated musculoskeletal symptoms among home care workers. Scand J Psych 36:113±129. Ê , Kadefors R, Rubenowitz S, Klingenstierna U, Johansson JA LindstroÈm I, EngstroÈm T, Johansson M. 1993. Musculoskeletal symptoms, ergonomic aspects and psychosocial factors in two different truck assembly concepts. Int J Ind Ergon 12:35±48. Ê , Rubenowitz S. 1994. Risk indicators in the psychosocial Johansson JA and physical work environment for work-related neck, shoulder and low back symptoms: a study among blue- and white-collar workers in eight companies. Scand J Rehab Med 26:131±142. Kamwendo K, Linton SJ, Moritz U. 1991. Neck and shoulder disorders in medical secretaries. Scand J Rehab Med 23:127±133. Ê , Persson J, Jonsson BG. 1986. Disorders of the cerviKilbom A cobrachial region among female workers in the electronics industry. Int J Ind Ergon 1:37±47. Koningsveld EAP, Mossink JCM, editors. 1997. Socio-economic costs of occupational safety and health in the Netherlands (in Dutch), VUGA Uitgeverij BV, `s Gravenhage. LagerstroÈm M, Wenemark M, Hagberg M, Hjelm EW. 1995. Occupational and individual factors related to musculoskeletal symptoms in ®ve body regions among Swedish nursing personnel. Int Arch Occup Environ Health 68:27±35. Lang JM, Rothman KJ, Cann CI. 1998. That confounded P-value. Epidemiol 9(1):7±8. Lau EMC, Sham A, Wong KC. 1996. The prevalence of and risk factors for neck pain in Hong Kong. J Pub Health Med 18:396±399. Linton SJ. 1990. Risk factors for neck and back pain in a working population in Sweden. Work Stress 4:41±49. MaÈkelaÈ M, HelioÈvaara M, Sievers K, Impivaara O, Knekt P, Aromaa A. 1991. Prevalence, determinants, and consequences of chronic neck pain in Finland. Am J Epidemiol 134:1356±1367.

Tharr D. 1995. Evaluation of work-related musculoskeletal disorders and job stress among teleservice center representatives. Appl Occup Environ Hyg 10:812±816. Toomingas A, Theorell T, MicheÂlsen H, Nordemar R. 1997. Associations between self-rated psychosocial work conditions and musculoskeletal symptoms and signs. Scand J Work Environ Health 23:130±139. Ursin H, Endresen IM, Ursin G. 1988. Psychological factors and self reports of muscle pain. Eur J Appl Physiol 57:282±290. van Tulder M, Assendelft WJJ, Koes BW, Bouter LM, and the editorial board of the Cochrane Collaboration Back Review Group. 1997. Method guidelines for systematic reviews in the Cochrane Collaboration back review group for spinal disorders. Spine 22:2323± 2330. Veiersted KB, Westgaard RH. 1993. Development of trapezius myalgia among female workers performing light manual work. Scand J Work Environ Health 19:277±283. Viikari-Juntura E, RiihimaÈki H, Tola S, Videman T, Mutanen P. 1994. Neck trouble in machine operating, dynamic physical work and sedentary work: a prospective study on occupational and individual risk factors. J Clin Epidemiol 47:1411±1422. Westgaard RH, Jansen T. 1992. Individual and work related factors associated with symptoms of musculoskeletal complaints. II Different risk factors among sewing machine operators. Br J Ind Med 49:154± 162. Westgaard RH, Jensen C, Hansen K. 1993. Individual and work-related risk factors associated with symptoms of musculoskeletal complaints. Int Arch Occup Environ Health 64:405±413. Zettenberg C, Forsberg A, Hansson E, Johansson H, Nielsen P, Danielsson B, Inge G, Olsson B-M. 1997. Neck and upper extremity problems in car assembly workers. A comparison of subjective complaints, work satisfaction, physical examination and gender. Int J Ind Ergon 19:277±289.